STATUS.md

Sol status

• Updated: 2026-07-28
• Production branch: master (normal Pages deployment consumes its successful CI SHA)
• Published crate: solar-ephemeris 0.2.0

This document reports implemented behavior. Historical design intent remains in WEB_REDESIGN_SPEC.md and SOLAR_SYSTEM_SPEC.md; current normative decisions are in SPEC.md, RFC_ALIGNMENT.md, and docs/adr/.

Current architecture

• Dependency-free static web application using native ES modules.
• Two audited Rust engines compiled to raw WebAssembly:
  • solar-wasm for deterministic reduced solar-surface simulation.
  • solar-ephemeris for local sky and solar-system calculations.
• Optional Python JPL Horizons/DE441 provider implementing the same provider-neutral ephemeris contract.
• Immutable, versioned JSON boundaries:
  • solar-state-snapshot.v2
  • ephemeris-snapshot.v2
  • system-snapshot.v1
• Real star catalogue behind every surface that draws stars: the naked-eye Hipparcos set (8,867 stars, V ≤ 6.5 — true J2000 positions, parallax distances, B−V colours) generated deterministically from committed sources by tools/generate_star_catalog.py and gated by tools/validate_star_catalog.py. The on-device engine reduces the 108-star bright subset with proper motion.
• Python generators and validators for deterministic fixtures, public-data normalization, schemas, semantic invariants, and external evidence.

Precision-hardening status

Solar engine

Implemented:

• Explicit west-positive heliographic Carrington coordinates.
• Explicit latitude-major, longitude-contiguous grid storage.
• Carrington-relative differential rotation.
• Partition-invariant Poisson event scheduling.
• Fixed-clock transport checkpoint/replay so target state does not depend on caller partitioning.
• Event-timed bipolar source injection.
• Exact exponential decay per integration segment.
• Strict finite/range checks before serialization.
• Readiness metadata derived from actual source and observation state.
• Complete semantic validation of all readiness gates.

Limits remain explicit:

• Magnetic fields are normalized, not calibrated Gauss/Mx.
• Diffusion is a tuned reduced flat-grid operator, not the exact spherical Laplacian.
• Meridional circulation is not implemented.
• No operational forecast authority is claimed.

Ephemeris engine

Implemented:

• Separate geocentric and topocentric apparent RA/Dec.
• Explicit UTC, TAI, TT, UT1, DUT1, leap-second, polar-motion, and EOP-quality metadata.
• Bundled IERS Bulletin A rapid/predicted EOP data with explicit degradation outside coverage.
• Release and weekly gates requiring at least 90 days of remaining EOP prediction coverage.
• VSOP2013/ELP-MPP02 apparent-place path and TOP2013 orbit-view support.
• JPL Horizons validation tooling.
• Body-specific rise/set thresholds:
  • planets/stars: standard refraction threshold;
  • Sun/Moon: refraction plus instantaneous apparent semidiameter after topocentric parallax.
• Rise/transit/set solved within the observer's local mean-solar day.
• Transit altitude reports the true topocentric centre altitude rather than the crossing margin.
• Nullable events where no event occurs; no fabricated values.

Accuracy scope remains explicit:

• Near-present apparent-place claims depend on current EOP coverage and the committed validation matrix.
• Deep-time heliocentric geometry does not imply deep-time topocentric pointing accuracy.
• Not for navigation, occultation prediction, mission safety, or safety-critical timing.

Provider continuity

Implemented:

• Local WASM and optional JPL server both emit ephemeris-snapshot.v2.
• The browser validates either provider through one runtime guard.
• Mixed v1/v2 responses are rejected.
• Geocentric/topocentric lunar aliasing is rejected.
• The optional server does not fabricate rise/transit/set values.
• The public web app no longer defaults to localhost for the remote provider.
• Remote provider controls are disabled unless a deployment explicitly configures an endpoint.
• First remote use requires consent to transmit selected latitude, longitude, elevation, and time.

Web and UX status

Implemented:

• Real NASA/SDO solar imagery with deterministic synthetic fallback.
• Sun-first newcomer path and progressively disclosed research details.
• Solar cycle playback and idealized latitude-vs-time butterfly diagram.
• Wavelength views, active-region inspection, space-weather learning context, and provenance/readiness display.
• My Sky observer view with geolocation/manual coordinates, time selection, share links, and JSON export.
• Solar System 3-D and top-down views rendered with WebGL2 (the earlier WebGPU path was retired during the orrery rewrite; see SOLAR_SYSTEM_SPEC P5 note).
• Milky-Way (galactic-scale) view with the Sun's orbit, differential-rotation shear, and deep-sky landmarks, plus a light-year-scale Solar neighbourhood sub-view placing the catalogue stars at their true parallax-derived 3-D positions around the Sun.
• Star rendering is catalogue-backed, not procedural: real positions, magnitudes, and B−V colours. Derived physics (luminosity, temperature, radius, and a labelled main-sequence mass estimate) lives in apps/web/js/starphysics.js.
• All 88 IAU constellation figures on both the 3-D sky and the My Sky dome, generated from committed IAU line data as RA/Dec polylines (tools/generate_constellations.py) rather than joined by star name.
• The 21 major moons of Mars, Jupiter, Saturn, Uranus and Neptune are drawn in their real orbits, lit and clickable. Orbits interpolate multi-epoch JPL Horizons osculating elements; tools/validate_moons.py gates the result in CI at worst 0.1384° angular and 0.1893% radial error across 11,985 interleaved, held-out checks. Satellite systems are inflated by one factor each so the inner moons clear their planet's disc and rings without distorting the spacing. This is a view, not a satellite ephemeris — not for occultations or mutual events, and the layer withholds itself rather than extrapolate: moons are drawn only inside the shared January 2021–December 2030 validation interval, and a moon is dropped when the animation clock outruns its orbit past the Nyquist rate. Both suppressions are explained in the accuracy line rather than left silent. Moons appear in the keyboard/screen-reader positions list, nested under their planet.
• Transiting moons cast their umbra/penumbra on the parent disc and darken inside its shadow cone, all computed in the planet's body frame from physical planetocentric positions with the spheroid handled exactly — display orbits are inflated, so the Overlays panel explains why the shadow does not sit under the drawn moon. Moon luminance is anchored to published geometric albedo (JPL Horizons physical-data blocks) through the sRGB transfer, within the browse products' residual stretch; tools/validate_body_motion.py gates every body's measured spin, sense, tilt and the moons' knot-implied periods in CI.
• Real surface geography ships with the repository independently of the committed photographic baseline: Earth's coastlines, lakes and permanent ice come from committed Natural Earth 1:110m vectors and the Moon's maria from the IAU/USGS gazetteer, generated by tools/generate_geography.py and rasterised in the browser by apps/web/js/surfacemap.js. Mars and Mercury deliberately stay procedural — the catalogue gives their features' positions and sizes but never their albedo, and a type-based guess gets Hellas Planitia backwards. Polygon interiors are honoured: GeoJSON inner rings subtract via an even-odd fill rather than being painted solid. The generated maps are independent of the "NASA textures" control, which governs only the photographic maps it names. A real photographic map from tools/fetch_textures.py still takes priority wherever present.
• Body orientation follows the full IAU WGCCRE convention, including the secular α0/δ0 rates that were previously omitted. Earth's axis precesses on its true cone rather than along the IAU's linear tangent, which is only valid near J2000 and would otherwise invert the prime meridian across the whole pre-J2000 half of the ±5000-year date slider.
• Clicking a body in the 3-D view opens a facts card built from apps/web/js/bodyData.js. The Moon's card states that its rotation is synchronous with its orbit and reports libration, so that the Moon visibly turning in the view reads as tidal locking rather than as a contradiction of it. Locking is emergent from the IAU rotation elements, not a stored flag, and tests/web/moonlock.test.mjs gates the sub-Earth point against winding.
• Named stars are clickable in both the sky and the neighbourhood view, opening a facts card that separates measured catalogue rows from derived ones, names the method for each, and withholds what it cannot honestly compute — everything distance-dependent when a star has no usable parallax, and the mass of an evolved star, where the main-sequence mass–luminosity relation does not apply.
• The ~370 KB catalogue module is lazy-loaded only when the 3-D view opens, so the Sun and My Sky first paint are unaffected — enforced by the @lazy-module rule in tools/validate_web_static.py.
• Keyboard-accessible region/body lists and canvas alternatives.
• Focus-trapped onboarding dialog with focus restoration.
• Reduced-motion CSS and 3-D auto-animation gating.
• Browser/device timezone disclosure for civil event times; UTC/JD remain authoritative in exports.
• Explicit remote-provider privacy disclosure.

Known product limitations:

• Observer IANA timezone is not inferred from coordinates; displayed civil times use the browser/device timezone.
• Deep-time topocentric precision is EOP/delta-T limited.
• Photographic textures are a committed baseline (planets, Earth's Moon, and ten catalogued-moon USGS mosaics); the eleven moons without a usable published global product render procedurally, each absence recorded with its reason in tools/fetch_textures.py's NO_MOSAIC.
• Deep-time axis orientation is first-order. Earth precesses on its true cone, but nutation, the slow change in obliquity, and the non-Earth bodies' linear IAU rates all remain approximations that soften toward the ends of the ±5000-year range.
• The remote DE441 provider requires a separately deployed endpoint.

Build and release integrity

Implemented:

• Specification-driven SDLC with 16 stable SOL-* requirements mapped to governing specifications, implementation, verification, and named CI gates.
• Internal RFC process/template, ADR alignment, contributor guide, and pull-request evidence checklist.
• Independent governance CI job validating requirement/RFC structure, evidence paths, dependency automation, immutable action pins, coverage thresholds, docs, UX structure, and exact-SHA deployment semantics.
• Rust workspace tests with locked dependencies.
• rustfmt and Clippy with warnings denied.
• WASM builds for both engines.
• Solar and ephemeris schema/semantic validation.
• Deterministic fixture and cycle-series regeneration.
• Offline local/server provider compatibility tests.
• EOP freshness gate.
• Native ES-module syntax checks.
• Built-WASM headless Chromium smoke tests for the Sun, My Sky, and 3-D Solar System paths.
• Static and real-browser progressive-disclosure assertions, including initial state, native keyboard toggling, destination state, panel state, and provider recovery.
• Denominator-complete Node + Chromium JavaScript line coverage at 90% or higher, with independent 90% Node line/branch/function gates.
• WebGL visual assertions for Sun colour, Earth visibility, and camera-orbit continuity.
• Immutable commit-SHA pins for external GitHub Actions.
• Pages deployment only after successful CI on master and only for the exact tested SHA.
• Procedural texture fallback by default; mutable remote texture fetching is reviewed opt-in behavior.

Repository-setting follow-up:

• Require CI, Coverage, and Docs checks and reviewed pull requests on master.
• Configure github-pages environment reviewers if the repository's risk policy requires human release approval.
• Approve security scope, supported versions, reporting ownership, severity context, and response expectations before adding a root SECURITY.md.
• Complete a scoped manual WCAG 2.2 AA audit before making any accessibility-conformance claim.

Release boundary

The application is research- and learning-ready. space_weather_operational remains false until all of the following exist:

1. Calibrated physical magnetic units.
2. Historical forecast validation and published skill evidence.
3. Comparison against operational SWPC products.
4. Adapter freshness monitoring and alerting.
5. Documented operational ownership, approval, and incident response.

No UI, snapshot, or deployment may represent Sol as an operational warning or mission-safety system before those gates pass.